Pruned Fuzzy Neural Networks
What is it about?
This article discusses a hybrid technique that uses concepts of artificial neural networks and fuzzy systems. As the creation of neurons made by the fuzzy method can generate solutions with several neurons, a pruning technique based on the concept of f-scores is used. In this methodology, binary pattern classification tests are applied.
Why is it important?
This model is important because it presents the union between the concepts of two techniques widely used in the artificial intelligence literature. In addition to uniting the interpretability of fuzzy systems and training provided by a neural network, this system uses pruning techniques of neurons in the hidden layer, allowing to simulate the selection of more relevant characteristics to obtain more precise answers. This model allows the creation of expert systems based on IF / THEN rules. This facilitates the use of people who are not present in daily AI studies, allowing expert systems to assist in solving binary problems.